Socio-Economic Perspectives on Shifting CultivationLandscapes in Northern Laos

Andreas Heinimann & Cornelia Hett & Kaspar Hurni &Peter Messerli & Michael Epprecht & Lars Jørgensen &

Thomas Breu

Published online: 22 January 2013# Springer Science+Business Media New York 2013

Abstract Despite the rapid agricultural transition that hasoccurred in the past decade, shifting cultivation remains awidespread agricultural practice in the northern uplands ofLao PDR. Little information is available on the basic socio-economic situation and respective possible patterns in shift-ing cultivation landscapes on a regional level. On the basisof a recent approximation of the extent of shifting cultiva-tion landscapes for two time periods and disaggregatedvillage level census data, this paper characterizes these

landscapes in terms of key socioeconomic parameters forthe whole of northern Laos. Results showed that over550,000 people live in shifting cultivation regions. Thepoverty rate of this population of 46.5 % is considerablyhigher than the national rural rate. Most shifting cultivationlandscapes are located in remote locations and a high shareof the population comprises ethnic minorities, pointing tomulti-dimensional marginality of these areas. We discusswhether economic growth and increased market accessibil-ity are sufficient to lift these landscapes out of poverty.

Keywords Shifting cultivation . Poverty . Accessibility .

Ethnicity . Lao PDR

Introduction

For centuries, shifting cultivation has been the dominantland use system in the north of the Lao Peoples DemocraticRepublic (Lao PDR; hereafter referred to as Laos), securingthe livelihoods of the large percentage of the rural popula-tion living in these mountainous regions. However, nowa-days, this traditional socio-ecological system is undergoinga rapid transformation. This transformation, from subsis-tence to market oriented agriculture production, has beenaccelerated by the increasing regional and global economicintegration together with the rapidly expanding major trans-portation infrastructure (Thongmanivong et al. 2009; Roderet al. 1997; Seidenberg et al. 2003; Padoch et al. 2007;World Bank 2008). In addition to these economic policiesand international market forces, other important drivingforces are the land use policies in general and those specificto shifting cultivation; since the mid-nineties the Lao gov-ernment has aimed to reduce the area under shifting culti-vation (Lestrelin et al. 2012). These policies are rooted inthe policy-makers’ general perception of shifting cultivation

A. Heinimann (*) : C. Hett : P. Messerli :M. Epprecht : T. BreuCentre for Development and Environment (CDE),University of Bern, Hallerstrasse 10,3012 Berne, Switzerlande-mail: andreas.heinimann@cde.unibe.ch

C. Hette-mail: cornelia.hett@cde.unibe.ch

P. Messerlie-mail: peter.messerli@cde.unibe.ch

T. Breue-mail: thomas.breu@cde.unibe.ch

K. Hurni : T. BreuSwiss National Centre of Competence in Research (NCCR)North-South, Institute of Geography, University of Bern,Hallerstrasse 10,3012 Berne, Switzerland

K. Hurnie-mail: kaspar.hurni@cde.unibe.ch

A. Heinimann :C. Hett :M. EpprechtNCCR North-South, Ministry of Environment and NaturalResource, Prime Minister’s Office compound,P. O. Box 1438, Vientiane, Laos

L. JørgensenDepartment of Geography and Geology,University of Copenhagen, Øster Voldgade 10,1350 Copenhagen K, Denmarke-mail: Lj@geo.ku.dk

Hum Ecol (2013) 41:51–62DOI 10.1007/s10745-013-9564-1

as a backward and underdeveloped form of land use (Mertzet al. 2009), representing a poverty trap (Bounthong et al.2003) and mainly responsible for the continuing deforesta-tion (Lawrence et al. 2010) and forest degradation (Fox2000; Thongmanivong et al. 2009). However, various localcase studies suggest that frequently, upland farmers havesimply not had many alternatives to shifting cultivation andthat these controversial policies have had little effect(Ducourtieux 2005; Roder 2001; Yokoyama 2004; Saito etal. 2006; Inoue et al. 2007; Leek 2007; Linquist et al. 2007;Yamamoto et al. 2009; Inoue et al. 2010; Hett et al. 2011;Alexander et al. 2010).

Since the current and future role of shifting cultivation inthe country’s agricultural development remains highly con-tested, the availability of reliable and systematic evidence onthe general characteristics and dynamics of this practicebeyond that of anecdotal insights of case studies is crucial.Currently, there are no national statistics on shifting cultiva-tion. It does not appear as a category in either the agricul-tural statistics or the socio-economic census. This iscommon throughout South East Asia (SEA) (Mertz et al.2009). Furthermore, since official national land cover inven-tories do not only fail to capture those areas under shiftingcultivation but, in fact, obscure them by classifying them as“potential forest” or “barren lands” (Hett et al. 2011), therewas, until recently, no established approach to the detectionof such land use across larger areas beyond that of local casestudies (Schmidt-Vogt et al. 2009). Hence, basic data suchas the areas under shifting cultivation or the populationliving in such regions are vague and very diverse (Mertz etal. 2009; Schmidt-Vogt et al. 2009).

This lack of regional level studies and consistent data hasresulted in intensive debates on the management of shiftingcultivation landscapes in Laos; debates that are usuallyhighly ideological but without any factual basis. The fol-lowing characteristics of shifting cultivation are commonlyassumed for the entity of the territory:

1) Shifting cultivation is mainly practiced in inaccessibleuplands,

2) Shifting cultivation is mainly practiced by the ethnicminority groups of the population and

3) Shifting cultivation regions are prone to high levels ofpoverty.

These assumptions must be considered as being onlyhypothetical in nature since no consistent or recent data,either for the country as a whole or even for large areas,exist to substantiate them.

Recent studies have made initial attempts to delineateshifting cultivation landscapes and estimate the percentagesof population involved. On the basis of a landscape mosaicapproach using existing land cover data from 2002, therecent study of Messerli et al. (2009) was able to identify

different dominant landscapes, including those under shift-ing cultivation at the national level. Using data from the2005 National Population and Housing Census, the authorswere also able to assess, for the first time, the populationliving in these landscapes. The study estimated that 17 % ofthe national population lived in the 29 % of the country thatcomprises shifting cultivation landscapes. More recently,Hurni et al. (2012, this issue) have developed an approachbased on remote sensing data, which estimates the shiftingcultivation dominated landscapes of northern Laos for twotime periods between 2000 and 2009. These studies havelaid the foundation for a further characterization of shiftingcultivation landscapes using selected socio-geographicindicators.

The aim of this paper is to shed light on the basic validityof the three above-mentioned general assumptionsconcerning shifting cultivation landscapes throughout north-ern Laos. Specifically we use the newly available spatiallyexplicit data on the extent of shifting cultivation areas innorthern Laos as well as socio-economic data at the villagelevel to analyze emerging (spatial) patterns in shifting cul-tivation dominated landscapes with regard to poverty, eth-nicity, accessibility as well as altitudinal belts. We hope onthe one hand, to create an empirical basis for more in-depthfuture analyses by experts from different disciplines and onthe other hand, to contribute with evidence of the generaldevelopments and policy debates on shifting cultivation inLaos.

Materials and Methods

Study Area

A considerable part of northern Laos is mountainous withlimited accessibility by a few all-season roads connectingonly the main district capitals. The traditional land usesystems are shifting cultivation (mainly upland rice) on theslopes and paddy rice cultivation in the few available valleybottoms. Over the past decade, increased regional economicintegration and policies promoting economic growth andinvestment have resulted in a swift transformation fromsubsistence to cash crop oriented agriculture and the intro-duction of tree plantations (Alton et al. 2005; Ducourtieux etal. 2006; Padoch et al. 2007; Cramb et al. 2009; Fox et al.2009; Thongmanivong et al. 2009).

The study region of the paper at hand is approximately158,000 km2 or two-thirds of the whole of Laos andincludes a large percentage of the shifting cultivation land-scapes initially estimated by Messerli et al. (2009). It wasnecessary to limit our study to this northern part of Laos asthe recent assessment of Hurni et al. (2012, this issue),which serves as one of our sources of baseline data, was

52 Hum Ecol (2013) 41:51–62

also limited to this region. For simplicity we hereafter referto this study area as “northern Laos”.

The Extent and Dynamics of Shifting CultivationLandscapes

Hurni et al. (2012, this issue) recently developed a newapproach using hyper-temporal MODIS satellite data inconjunction with multi-temporal Landsat imagery from2000 to 2009 to detect shifting cultivation landscapes. Thisapproach is based on the dominance of the very specificspatio-temporal land cover pattern of clearings and regrowthleft on the landscape by shifting cultivation. In their re-search, Hurni et al. (2012, this issue) estimated the extentof shifting cultivation dominated landscapes (hereaftercalled shifting cultivation landscapes) for northern Laosfor two time periods: 2000–2006 and 2003–2009. Thisoverlap in time was necessary as 6 years of data wereneeded to delineate these landscapes. The assessment doesnot differentiate the land use systems or land cover outsideof the delineated shifting cultivation landscapes. For thestudy region these are mainly forests, permanent agriculturallandscapes and some built-up and water areas. This has to beconsidered when interpreting any comparison of socio-economic characteristics of shifting cultivation landscapeswith those of non-shifting cultivation areas, as presented inthe study at hand.

Data on the Socio-Economic Setting

The 2005 Population and Housing Census of the Lao De-partment of Statistics is the only coherent and spatiallydisaggregated countrywide socio-economic baseline in Laos(Messerli et al. 2008). We used village level aggregates ofthis census data for all of the 6,521 villages (62 % of allvillages in Laos and 58 % of the Lao population) in ourstudy area. For data on poverty, we used the village levelpoverty estimates of Epprecht et al. (2008), which are esti-mates of the number of people living below the nationalpoverty line. This estimation was based on the ‘small areaestimation’ method and combined information derived fromthe 2005 Population and Housing Census with informationfrom the Lao Expenditure and Consumption Survey (LECS)of 2003. The poverty measure used by Epprecht at al. (2008)uses a welfare indicator of the value of per capita consump-tion expenditure, including the value of subsistence foodproduction and the imputed rental value of owner occupiedhousing. The poverty line used there corresponds to the percapita expenditure, including the value of home production,required to purchase 2,100 Kcal per person per day usingthe food basket of households in the third quintile, plus anon-food allowance equal to what households in the thirdquintile spend on non-food items. In this sense, poor people

are not at an adequate subsistence level, as their per capitaconsumption expenditure is not enough to cover basic needssuch as minimal health care, education or food. This has tobe kept in mind when interpreting the results of this study asthere are a number of other poverty measurementapproaches that take into account an even wider conceptof well-being. For Laos, however, the expenditure basedpoverty measure of Epprecht et al. (2008) provides the onlyavailable national data at village level and was thereforeused in this study.

Furthermore, the use of socio-economic data for only onetime period, aggregated at the village level, has two impor-tant consequences for the interpretation of results: 1) All thechanges in socio-economic characteristics of shifting culti-vation landscapes presented were based on a change in theland use pattern over time and not on the socio-economicdata. 2) It was not possible to reach any conclusion on thecharacteristics of sub-village populations. Therefore, al-though this study provided, for example, information onthe total number of people living within shifting cultivatedlandscapes, not all villagers within this number may dependexclusively on shifting cultivation.

Data on Accessibility

The 2005 Population Census does not provide any informa-tion (beyond simple indicators of road access) on villageremoteness. Accessibility, here understood as travel time toland itself and to basic services such as markets, has beenshown to be a key determinant of land use and land usechanges (Angelsen and Kaimowitz 1999; Castella et al.2005; Chomitz and Gray 1996; Geist and Lambin 2002;Verburg et al. 2004; Epprecht et al. 2010).

This is especially relevant in Laos, as the still very weaktransportation infrastructure in general is currently beingupgraded through massive investments in this sector; firstly,through mega projects such as the establishment of economicdevelopment corridors in the Greater Mekong Sub-region(GMS) by the Asian Development Bank (ABD), or morerecently, through the discussion of a Chinese funded high-speed railroad though Laos, and secondly, through a slightlyslower process of upgrading smaller roads to provide ruralareas with access to these economic development corridors.As a result, rural accessibility is being enhanced quite swiftlyin Laos.

In order to provide insights into possible basic patterns ofaccessibility related to shifting cultivation landscapes, a modelof physical accessibility has been developed. This modelestimates travel time from any point in the study area to, forexample, the nearest district capital using standard GIS rasterbased (in our case, 100 m resolution) cost-distance functions(see Fig. 1). District capitals were chosen as the “sources” (ordestinations) for this model, as they frequently act as a

Hum Ecol (2013) 41:51–62 53

development gateway or entry point to the more remote hin-terlands (Messerli et al. 2008). Thus this accessibility modelmay also be partly understood as yielding an approximation ofmarket accessibility. When interpreting the accessibility relat-ed result of this study, the following has to be kept in mind:firstly, physical accessibility does not automatically imply

actual access to services. In this study we only considerphysical accessibility and not socio-economic distance asdefined e.g. by Epprecht et al. (2010). Secondly, our accessi-bility model estimates travel time to certain destinations,taking into account the availability and quality of road net-works, terrain, land cover and potential means of transport—

54 Hum Ecol (2013) 41:51–62

Fig. 1 District accessibility: Travel time from and to district capitals

thereby assuming access to typically available means oftransport.

Overview of Approach

The different datasets employed in this study vary interms of their basic geometries or spatial units of refer-ence. While the village level census data use an estimat-ed village area (i.e. the so-called village polygon) as areference, the assessment of Hurni et al. (2012, thisissue) used the delineated borders of shifting cultivationlandscapes, and the accessibility data are raster based. Tohave a spatial unit of analysis at hand for this study, weused standard spatial intersections of the village polygonswith the data on the extent of shifting cultivation for thetwo time periods as new basic geometries for all furtheranalyses. For the assignment of the socio-economicattributes based on the village geometries to the newpolygons created through the spatial overlay we followedan area-based approach: the absolute data of the socio-economic variables (e.g. population, ethnicity) were at-tributed to the new geometries on the basis of the re-spective share of area of the village polygon that wasdominated by shifting cultivation in one of the two timeperiods considered. This mitigates to a certain extent theissue of using only village level aggregated data (seeabove) but naturally cannot solve it.

In order to depict socio-economic characteristics of shift-ing cultivation landscapes spatially at one point in time andprovide insights into the larger overall patterns of povertyand accessibility, a landscape typology was developed. Thistypology was based on different combinations of the fol-lowing three main factors:

1) Presence of shifting cultivation landscapes in the sec-ond time period considered (2003–2009; based onHurni et al. 2012, this issue).

2) The travel time to district capitals based on the accessi-bility model. This was classified as two spatial categories,i.e. regions from which a district capital is reachablewithin a day trip (<2 h) and regions from which a districtcapital is more than a day trip away (>2 h). Hence a “daytrip” is defined as a four-hour travel time (there and back),which we consider as a reasonable estimate in order toreach markets and services such as agricultural extensionservices.

3) The poverty data used were reclassified in two classes:areas where village-level poverty rates were lower thanthe national rural poverty rate (hereafter referred to as“better off”) and those where rates were higher (hereaf-ter referred to as “poorer”). The latest countrywideestimate of the national rural poverty rate of 40 % wasused (Epprecht et al. 2008).

Results

The following results section begins with an overview of thepopulation and poverty figures in shifting cultivation land-scapes, followed by the pattern of the distribution of shiftingcultivation landscapes across travel times to district capitals,the ranges of elevation and the composition of ethno-linguistic families, and ends with the presentation of atypology of shifting cultivation landscapes in relation toaccessibility and poverty.

Estimation of Numbers and Poverty Rates of the PopulationLiving in Shifting Cultivation Landscapes

Using the recent delineation of shifting cultivation land-scapes of Hurni et al. (2012, this issue), the latest availablecensus information (Messerli et al. 2008) and poverty esti-mates (Epprecht et al. 2008) we were able to estimate thepopulation and the respective poverty rates in shifting cul-tivation landscapes in northern Laos in two time periods (seeTable 1). In northern Laos, both the absolute numbers andthe percentages of the population living in shifting cultiva-tion landscapes were found to be considerable. Over550,000 people or 15.6 % of the total population lived insuch landscapes in the study area (2003–2006).

In the time periods examined (2000–2006 and 2003–2009)the area of shifting cultivationed landscapes decreased by3,500 km2 (or 10 %) while the population living in suchlandscapes decreased by 115,300 people (or 17.3 %).

Poverty rates in shifting cultivation landscapes were45.6 % in 2000–2006 and 46.5 % in 2003–2009. This slightincrease between the two time periods (see Table 1) does notimply a change in poverty per se, but rather points to the factthat the shifting cultivation landscapes in the second timeperiod were located in poorer regions. However, both theserates were considerably higher than the national poverty rate(37.7 %) of the total population in 2005 and that of the ruralpopulation (40 %) (Epprecht et al. 2008).

Accessibility and Areas of Shifting Cultivation Landscapes

A clear pattern of the distribution of shifting cultivationlandscapes across travel times to district capitals emerged(see Fig. 2). Only small percentages (8 % in 2000–2006 and6 % in 2003–2009) of the total of shifting cultivation land-scapes were found in the immediate vicinity of districtcapitals while percentages increased dramatically in areasslightly further away. A further decrease in accessibilitycorrelated with a gradual decrease in areas under shiftingcultivation. The increase in regions >8 h must be interpretedwith care as the respective accessibility class does not fol-low the 1 h classification system and hence comprises thecumulative shifting cultivation areas beyond 8 h. Overall,

Hum Ecol (2013) 41:51–62 55

the largest shares of shifting cultivation landscapes werefound in regions beyond a 2 h travel time (71.4 % and72.6 % of the total respective area of the two time periodsconsidered). Hence, there is a slight tendency for the shiftingcultivation landscapes to be located in less accessibleregions over time. In contrast, percentages of shifting culti-vation landscapes in the vicinity of towns sharply decreasedbetween the two time periods and more than 33 % of allshifting cultivation landscapes closer than 1 h to districtcapitals disappeared. This decrease accounted for over27 % of the total net decrease in shifting cultivation land-scapes between the two time periods.

When the entire study area was considered, shifting cul-tivation landscapes proved not to be the dominant landscapeat any accessibility level (see Fig. 3). However, it is stillinteresting to note that except for the most accessible and themost remote areas, shifting cultivation landscapes representa very important landscape element across all other acces-sibility classes.

Furthermore, and linked to accessibility, is the questionof the range of elevations at which shifting cultivation land-scapes can be found. As already mentioned, shifting culti-vation is commonly assumed to be an upland agriculturesystem in Laos. Our results clearly show that this is indeed

the case. Over 78 % of the shifting cultivation landscapes in2003–2009 were above 600 masl, which can be consideredupland regions in a Lao context (see Table 2). The generalpattern across the elevation range confirms the intuitiveassumption: while only a small share of shifting cultivationis located in the lowlands (due to the dominance of e.g.permanent agriculture) the majority is located at intermedi-ate to high elevation. The share decreases in very highelevation ranges due to rough terrains and some large plainswith natural grasslands (e.g. Xieng Khouang plain).

Ethno-Linguistic Families Living in Shifting CultivationLandscapes

There are 49 different ethnic groups in Laos. These areusually classified as belonging to four ethno-linguistic fam-ilies: Lao-Tai (Tai-Kadai); Mon-Khmer (Austro-Asiatic);Hmong-Mien (Hmong-Yao, Miao-Yao); and Sino-Tibetan(mostly Tibeto-Burman) (Messerli et al. 2008). The Lao-Tai ethno-linguistic family is dominant at the national level(comprising 65 % of the total population) and also in ourstudy area (comprising 60 %), while the other groups can, tosome degree, be considered a collection of ethnic minorities(Messerli et al. 2008).

Table 1 Basic socio-economicinformation on shifting cultiva-tion landscapes in northern Laosin 2000–2006 and 2003–2009(area estimates are based onHurni et al. (2012, this issue))

Shifting cultivationlandscapes 2000–2006 (t0)

Shifting cultivationlandscapes 2003–2009 (t1)

Net-change ofshifting cultivationlandscapes t0–t1

Area km2 35,140 31,640 −3,500

Percentage of totalstudy area

22.2 20.0 −2.2

Population 667,300 552,000 −115,300

Percentage of populationof total study area

19.0 15.6 −3.3

Poverty rate 45.6 46.5 –

Fig. 2 Distribution of shiftingcultivation landscapes acrossaccessibility to district capitals

56 Hum Ecol (2013) 41:51–62

However, in shifting cultivation landscapes in northernLaos (2003–2009) the Lao-Tai made up only 19.4 % of thetotal population while the largest percentage was composed ofother families (Mon-Khmer: 55.0 %, Sino-Tibetan: 9.8 % andHmong-Mien: 15.8 %) (see Fig. 4). This basically confirms theinformation found in local case studies (e.g. Thongmanivonget al. 2009; Fujita and Phanvilay 2008; Ducourtieux et al.2006; Thongmanivong and Fujita 2006) and provided theevidence for the whole of northern Laos of shifting cultivationlandscapes being mainly inhabited by ethnic minorities. Forexample, while only 5 % of the total Lao-Tai population of thestudy area lives in shifting cultivation landscapes the area ishome to almost 33 % of the entire population of the other threeethno-linguistic families. This emphasizes the importance ofshifting cultivation landscapes for the lives of all of the ethnicminority families in the area. The clear national pattern is thatthe Lao-Tai live mainly in the easily accessible areas in thefloodplains (Messerli et al. 2008). This pattern was also ob-servable in our results, even though we considered only thoseLao-Tai living in shifting cultivation landscapes in northernLaos, wheremore than 70% of the population lived closer than2 hours’ travel from a district capital, while this was true forsignificantly less than 50 % of the other groups.

There was no change in this overall pattern of shiftingcultivation landscapes in the two time periods considered. Inshifting cultivation landscapes, the percentage of Lao-Tai of

the total population did, however, decrease slightly from21.1 to 19.4 %. Hence there is a trend that the shares ofthe other three ethno-linguistic families in shifting cultiva-tion landscapes are still increasing.

Detecting Patterns: Landscape Typology of ShiftingCultivation Landscapes in 2003–2009

In order to be able to depict the basic socio-economiccharacteristics of shifting cultivation landscapes spatiallyin the first time period considered (2003–2006) and toobtain insights into the larger overall patterns of povertyand accessibility in shifting cultivation landscapes, a land-scape typology was developed and is presented here. Thislandscape typology classifies the shifting cultivation land-scapes by using the two main criteria of poverty andaccessibility.

More than 75 % of the shifting cultivation landscapes arein regions that are poorer than the national rural average (seeTable 3). These areas are also home to more than 66 % ofthe population of the study region. This does not imply thatall the people are poor, but rather that they live in a land-scape where the majority of people are poor, i.e. with apoverty rate higher than the national rural poverty rate.The actual calculated poverty rate in shifting cultivationlandscapes (2003–2009) was 46.5 % (see Table 1).

Fig. 3 Shares of shiftingcultivation landscapes (2003–2009) of the entire study area inrelation to accessibility

Table 2 Percentages of shifting cultivation and all other landscapes in 2003–2009 per elevation zones

Elevation zones [meter asl]

0–200 201–400 401–600 601–800 801–1000 1001–1200 1201–1400 1401–1600 >1601 Total

Shifting cultivation landscapes 0.4 % 6.7 % 14.9 % 33.3 % 30.3 % 10.8 % 3.4 % 0.4 % 0.0 % 100 %

Not shifting cultivation landscapes 4.7 % 11.5 % 14.8 % 19.5 % 20.6 % 18.3 % 7.9 % 2.1 % 0.5 % 100 %

Hum Ecol (2013) 41:51–62 57

A further sub-classification of the landscapes into areasbeyond and within a day trip reveals clear patterns (seeTable 3). In shifting cultivation landscapes where villagepoverty rates are lower than the national rural poverty rate,the highest percentage (72.8 %) of the population lived withina day trip of a district capital. This situation ismainly related tothe higher population densities in these areas, since the per-centages of areas within (50.4 %) and beyond (49.6 %) a daytrip were almost equal. In regions with poverty rates higherthan the national rural poverty rate, most people (68.1 %) werefound to live in those areas beyond a day trip and the respec-tive share of area was equally dominant (78.5 %).

In terms of the ethno-linguistic composition of the fourlandscape typologies (see Table 3), it is striking that in poorerlandscapes, almost 70 % of the population who lived bothwithin (69.3 %) and beyond (68.5 %) a day trip are from theMon-Khmer family. In contrast, the chances of “finding”mem-bers of the national ethnic majority (Lao-Tai) in such land-scapes were shown to be very small (6.6 % and 9.1 % withinand beyond a day trip, respectively). Yet, in better off

landscapes the Lao-Tai make up more than 40 % of the popu-lation living within (48.2 %) and beyond (42.3 %) a day trip.

Figure 5 depicts the resulting spatial pattern of the typologyof shifting cultivation landscapes in 2003–2009. Shifting cul-tivation landscapes that have a higher poverty rate (“poorer”)than the national rural poverty and are beyond a day triprepresent the largest share (59.1 % of area and 45.1 % ofpopulation) and are mainly large connected patches. A certainclustering of better off shifting cultivation landscapes werefound in the north-west (in Luang Namtha Province) and inthe very north (in Phongsaly Province) of Laos.

Discussion

In the following discussion we briefly reflect on how theapproach chosen in this paper compares to others in terms ofthe most important results. We will then discuss the com-mon beliefs and assumptions about shifting cultivation inLaos in order to present a more differentiated picture.

Fig. 4 Percentages of thepopulation of the four ethno-linguistic families in shiftingcultivation landscape (2003–2009) and all other landscapesin the study area

Table 3 Results of the landscape typology of shifting cultivation landscapes (2003–2009) by poverty and accessibility

Shifting cultivation (SC) landscapes

Poverty SC landscapes with lower poverty rate(“better off”) than the national ruralpoverty rate

SC landscapes with higher poverty rate(“poorer”) than the national ruralpoverty rate

Percentage of area 24.6 75.4

Percentage of total population per landscape type 33.8 66.2

Accessibility to district capital Within daytrip Beyond daytrip Within daytrip Beyond daytrip

Percentage of above area 50.4 49.6 21.5 78.5

Percentage of above population 72.8 27.2 31.9 68.1

Ethnic composition of population of each poverty/accessibility class

Percentage of Lao-Tai 48.2 42.3 6.6 9.1

Percentage of Mon-Khmer 35.0 24.8 69.3 68.5

Percentage of Sino-Tibetan 5.8 23.4 4.1 6.7

Percentage of Hmong-Mien 11.1 9.5 20.0 15.7

Total ethnic composition 100 100 100 100

58 Hum Ecol (2013) 41:51–62

Finally, we will explore the relevance of these insights forthe current debates on development policy in the country.

It is difficult to make a direct comparison with earlier(e.g. Fujisaka 1991) and more recent (Messerli et al. 2009)attempts at estimating the extent and the population ofshifting cultivation landscapes. These estimations were

either rather vague (Mertz et al. 2009) or, as was the casein the estimations of Messerli et al. (2009), employed meth-odologies not directly comparable to ours. Nevertheless, acautious comparison of the national figures of Messerli et al.(2009) with our figures revealed a certain similarity, al-though these figures of 29 % of the area and 17 % of the

Fig. 5 Map of typology of shifting cultivation landscapes (2003–2009) and the percentage of area and population in northern Laos

Hum Ecol (2013) 41:51–62 59

population of shifting cultivation landscapes are slightlyhigher than ours. The different time periods of the observa-tions could well explain the small deviation and we maytentatively conclude that the main results of our study couldbe confirmed by two different approaches.

Our insights provide a base for reflecting on the wide-spread beliefs and generalisations that have come from localcase studies on shifting cultivation in Laos. Firstly, we canconfirm that shifting cultivation areas can indeed be charac-terized by considerably higher poverty rates than the nation-al average in rural areas. Secondly, shifting cultivation isgenerally found in remote upland areas. More specifically,the largest shares of shifting cultivation landscapes werefound in regions beyond 2 hours’ travel from district capi-tals. Furthermore, shifting cultivation landscapes are mainly(78 %) found above 600 masl, corresponding to the uplandsLaos. Thirdly, in such landscapes the Mon-Khmer ethno-linguistic family, an ethnic minority group at the nationallevel, predominates. The percentages of other minority pop-ulations are also elevated in shifting cultivation landscapeswhen compared to national population statistics, while theoverall national majority, the Lao-Tai, account for less thanone fifth of the population in shifting cultivation areas.

Despite a general correspondence of our results with thealready mentioned general assumptions on shifting cultivationwe believe that a more differentiated picture needs to be drawnin terms of poverty, remoteness and ethnic composition. Ifaccessibility to district capitals is considered a proxy indicatorfor people’s access to markets it is questionable whethershifting cultivation landscapes — given their dominance inremote areas— will benefit from a stronger market economy.This contradicts the transition, currently being promoted, froma planned to a market economy, whereby the government isattempting to incorporate even the remotest rural areasthrough economic growth, increased food security and greateropportunities for people’s participation (GOL 2006).

More precisely, the general remoteness of shifting culti-vation landscapes confirmed through our research implies alack of access not only to markets and capital, but also toother services such as agriculture extension and health serv-ices, and to information and technology. This greatly limitsthe options for future agricultural development in theseremote locations (Alexander et al. 2010), making shiftingcultivation farmers highly dependent on forest resources fortheir livelihoods (Cavendish 2003).

Finally, we would also like to draw attention to the resultsshowing that ethnicity plays a key role in terms of poverty inshifting cultivation landscapes, independent of their acces-sibility. While high poverty rates can be related to highpercentages of ethnic minorities, we observed recurrentpatterns of shifting cultivation where lower poverty ratesalso coincided with high percentages of the ethnic minoritypopulations. This may be interpreted as showing that it may

be difficult for certain ethnic minorities engaged in shiftingcultivation to escape poverty even if they are living ingeographically accessible areas. In other words, social dis-tances or marginality may be more important than physicaldistances to markets, services, infrastructure, etc. This hy-pothesis can be further strengthened by our results indicat-ing that the percentage of marginal shifting cultivationlandscapes has increased over time.

This issue is certainly relevant in terms of the highlydebated resettlement policy, a key instrument of thegovernment of Laos for upland development and povertyalleviation. It has been a longstanding policy of the Laogovernment to integrate these remote populations oflargely ethnic minorities into the ethnic lowland society(Evrard and Goudineau 2004; Ireson and Ireson 1991).The initial argument for such integration was that ofnational security. This was replaced by one focusing ona poverty alleviation strategy that involved facilitatingstate service delivery and market access (Lestrelin et al.2012; Baird and Shoemaker 2007). More recently, shift-ing cultivation has been blamed for causing environmen-tal degradation (e.g. deforestation, soil erosion) and thiswas used as the rationale for the village resettlementpolicy (Lestrelin 2010; Fox 2000). Our results provideevidence that these policies, targeting the resettlement ofmarginal shifting cultivation villages, may have littleeffect, as improved accessibility may not be enough toalleviate the poverty of ethnic minorities (Alexander etal. 2010; Ducourtieux 2005). Therefore, shifting cultiva-tion landscapes persist as the dominant land use in re-mote areas of northern Laos.

Conclusion and Outlook

In this study we have presented selected socio-economiccharacteristics of shifting cultivation landscapes and visual-ized their respective patterns for the whole of northern Laos.We have shown that a large share of the shifting cultivationlandscapes in northern Laos were indeed marginal fromvarious perspectives, i.e. they are in remote areas, wherepoverty rates are higher than the national rural poverty rateand the percentages of ethnic minorities are disproportion-ately high compared to country-wide statistics. These resultsconfirm, for the first time, for the whole of northern Laos thevalidity of similar insights gained from single case studies.

The results of this study on shifting cultivation land-scapes at an aggregated level for the whole of northern Laosare intended to contribute not only to the current debates ondevelopment policy but also to scientific discourses on thedynamics of shifting cultivation. At the same time, ouranalysis gives rise to new questions and opens the door tofurther in-depth studies. Additional case studies should

60 Hum Ecol (2013) 41:51–62

disentangle the complex interactions between remoteness,ethnicity and poverty in a shifting cultivation context. Fur-thermore, the understanding of changes in shifting cultiva-tion landscapes could be further improved by conducting acomparative analysis with the dynamics of other existingland use types. This would make it possible to reveal pos-sible trade-offs in terms of the different economic, socialand cultural functions of land that the ongoing transforma-tion processes bring about.

Acknowledgments We acknowledge support from the Swiss Nation-al Centre of Competence in Research (NCCR) North-South, co-fundedby the Swiss National Science Foundation (SNF), the Swiss Agencyfor Development and Cooperation (SDC) and the Centre for Develop-ment and Environment, University of Bern, Switzerland. Furthermore,this research has benefited from collaboration and support of the I-REDD+ project by the European Community’s Seventh FrameworkResearch Programme, the Agro Biodiversity Initiative (TABI) in Laos,as well as the Global Land Project (GLP).

References

Alexander, K. S., Millar, J., and Lipscombe, N. (2010). SustainableDevelopment in the Uplands of Lao PDR. Sustainable Develop-ment 18(1): 62–70.

Alton, C., Blum, D., and Sananikone, S. (2005). Para Rubber Study:Hevea Brasiliensis Lao PDR. GTZ, Lao German Program RuralDevelopment in Mountainous Areas of Northern Lao PDR. Vien-tiane, Lao PDR.

Angelsen, A., and Kaimowitz, D. (1999). Rethinking the Causes ofDeforestation: Lessons from Economic Models. The World BankObserver 14: 73–98.

Baird, I. G., and Shoemaker, B. (2007). Unsettling Experiences: Inter-nal Resettlement and International Aid Agencies in Laos Devel-opment and Change 38(5): 865–888.

Bounthong, B., Raintree, J., and Douangsavanh, L. (2003). Uplandagriculture and forestry research for improving livelihoods in LaoPDR. In Saxena, K. G., Liang, L., Kono, Y., and Miyata, S. (eds.).Environment and Sustainable Development. Proceedings of aninternational symposium, The United Nations University, Tokyo,Japan, 29–30 October, 2003. Small-scale livelihoods and naturalresources management in marginal areas: Case studies in mon-soon Asia. UNU Press, Tokyo.

Castella, J.-C., Manh, P. H., Kam, S. P., Villano, L., and Tronche, N. R.(2005). Analysis of Village Accessibility and Its Impact on LandUse Dynamics in a Mountainous Province of Northern Vietnam.Applied Geography 25(4): 308–326.

Cavendish, W. (2003). How do Forests Support, Insure and Improvethe Livelihoods of the Rural Poor? A research note. CIFOR.

Chomitz, K. M., and Gray, D. A. (1996). Roads, Land Use, andDeforestation: A Spatial Model Applied to Belize. The WorldBank Economic Review 10: 487–512.

Cramb, R., Colfer, C., Dressler,W., Laungaramsri, P., Le, Q.,Mulyoutami,E., Peluso, N., and Wadley, R. (2009). Swidden Transformationsand Rural Livelihoods in Southeast Asia. Human Ecology 37(3):323–346.

Ducourtieux, O. (2005). Shifting cultivation and poverty eradication: Acomplex issue. In Bouahom, B., Glendinning, A., Nilsson, S., andVictor, M. (eds.). Poverty Reduction and Shifting Cultivation Stabi-lisation in the Uplands of Lao PDR: Technologies, Approaches andMethods for Improving Upland Livelihoods. Proceedings of a

workshop held in Luang Prabang, Lao PDR, January 27–30, 2004.National Agriculture and Forestry Reserach Institute. Vientiane, LaoPDR. Pp. 71–94.

Ducourtieux, O., Visonnavong, P., and Rossard, J. (2006). IntroducingCash Crops in Shifting Cultivation Regions — The Experiencewith Cardamom in Laos. Agroforestry Systems 66(1): 65–76.

Epprecht, M., Minot, N., Messerli, P., and Heinimann, A. (2008).Poverty and Inequality in the Lao PDR: Spatial Patterns andGeographic Determinants. Geographica Bernensis, Bern.

Epprecht, M., Heinimann, A., Messerli, P., Schmidt-Vogt, D., andWiesmann, U. (2010). Accessibility as a determinant of environ-mental dynamics and socio-economic disparities in mainland South-east Asia. In: Hurni, H., and Wiesmann, U. (eds. with aninternational group of co-editors). Global Change and SustainableDevelopment: A Synthesis of Regional Experiences from ResearchPartnerships. Perspectives of the Swiss National Centre of Compe-tence in Research (NCCR) North-South, University of Bern, Vol. 5.Bern, Switzerland: Geographica Bernensia, pp 371–384.

Evrard, O., and Goudineau, Y. (2004). Planned Resettlement, Unex-pected Migrations and Cultural Trauma in Laos. Developmentand Change 35(5): 937–962.

Fox, J. M. (2000). How Blaming ‘Slash and Burn’ Farmers is Defor-esting Mainland Southeast Asia. AsiaPacific Issues 47. East-WestCenter, Honolulu.

Fox, J., Fujita, Y., Ngidang, D., Peluso, N., Potter, L., Sakuntaladewi,N., Sturgeon, J., and Thomas, D. (2009). Policies, Political-Economy, and Swidden in Southeast Asia. Human Ecology 37(3): 305–322.

Fujisaka, S. (1991). A Diagnostic Survey of Shifting Cultivation inNorthern Laos: Targeting Research to Improve Sustainability andProductivity. Agroforestry Systems 13: 95–109.

Fujita, Y., and Phanvilay, K. (2008). Land and Forest Allocation in LaoPeople’s Democratic Republic: Comparison of Case Studies fromCommunity-Based Natural Resource Management Research. So-ciety and Natural Resources 21(2): 120–133.

Geist, J. J., and Lambin, E. F. (2002). Proximate Causes and UnderlyingDriving Forces of Tropical Deforestation. Bioscience 52(2): 143–150.

GoL (2006). Strategic Plan on Governance (2006 – 2010). The Gov-ernment of the Lao PDR. November 2006. Vientiane, Lao PDR.

Hett, C., Castella, J. C., Heinimann, A., and Messerli, P. (2011). ALandscape Mosaics Approach for Characterizing Swidden Sys-tems from a REDD+. Applied Geography 32: 608–618.

Hurni, K., Hett, C., Heinimann, A., Messerli, P., and Wiesmann, U.(2012). Dynamics of Shifting Cultivation Landscapes in NorthernLaos Based on an Analysis of MODIS Time Series and LandsatImagery Between 2000 and 2009. Human Ecology doi:10.1007/s10745-012-9551-y.

Inoue, Y., Qi, J., Olioso, A., Kiyono, Y., Horie, T., Asai, H., Saito, K.,Ochiai, Y., Shiraiwa, T., and Douangsavanh, L. (2007). Traceabil-ity of slash-and-burn land-use history using optical satellite sensorimagery: a basis for chronosequential assessment of ecosystemcarbon stock in Laos. International Journal of Remote Sensing 28(24): 5641–5647.

Inoue, Y., Kiyono, Y., Asai, H., Ochiai, Y., Qi, J., Olioso, A., Shiraiwa,T., Horie, T., Saito, K., and Dounagsavanh, L. (2010). AssessingLand-Use and Carbon Stock in Slash-and-Burn Ecosystems inTropical Mountain of Laos Based on Time-Series SatelliteImages. International Journal of Applied Earth Observation andGeoinformation 12(4): 287–297.

Ireson, C., and Ireson, R. (1991). Ethnicity and Development in Laos.Asian Survey 31(10): 920–937.

Lawrence, D., Radel, C., Tully, K., Schmook, B., and Schneider,L. (2010). Untangling a Decline in Tropical Forest Resil-ience: Constraints on the Sustainability of Shifting Cultiva-tion Across the Globe. Biotropica 42(1): 21–30 doi:10.1111/j.1744-7429.2009.00599.x.

Hum Ecol (2013) 41:51–62 61

Leek, K. E. (2007). Rural Livelihood Strategies and Natural Resources inOudomxay, Lao PDR [MSc Thesis]. University of Natural Resourcesand Applied Life Sciences, Vienna; Lincoln University, Christchurch,NZ, Natural Resources Management and Ecological Engineering.

Lestrelin, G. (2010). Land Degradation in the Lao PDR: Discoursesand Policy. Land Use Policy 27: 424–439.

Lestrelin, G., Castella J.-C., and Bourgoin, J. (2012). TerritorialisingSustainable Development: The Politics of Land-Use Planning inthe Lao People’s Democratic Republic. Journal of ContemporaryAsia 42(4): 581–602.

Linquist, B., Trösch, K., Pandey, S., Phouynyavong, K., and Guenat,D. (2007). Montane Paddy Rice: Development and Effects onFood Security and Livelihood Activities of Highland Lao Farm-ers. Mountain Research and Development 27(1): 40–47.

Mertz, O., Leisz, S. J., Heinimann, A., Rerkasem, K., Thiha, T.,Dressler, W., Pham, V. C., Vu, K. C., Schmidt-Vogt, D., Colfer,C. J. P., Epprecht, M., Padoch, C., and Potter, L. (2009). WhoCounts? Demography of Swidden Cultivators in Southeast Asia.Human Ecology 37: 281–289 doi:10.1007/s10745-009-9249-y.

Messerli, P., Heinimann, A., Epprecht, M., Phonesaly, S., Thiraka, C.,and Minot, N. (eds.) (2008). Socio-Economic Atlas of the LaoPDR: An analysis based on the 2005 Population and HousingCensus. Swiss National Center of Competence in Research(NCCR) North-South, University of Bern, Bern and Vientiane.Geographica Bernensia.

Messerli, P., Heinimann, A., and Epprecht, M. (2009). Finding Homo-geneity in Heterogeneity—A New Approach to QuantifyingLandscape Mosaics Developed for the Lao PDR. Human Ecology37: 291–304 doi:10.1007/s10745-009-9238-1.

Padoch, C., Coffey, K., Mertz, O., Leisz, S. J., Fox, J., and Wadley, R.L. (2007). The Demise of Swidden in Southeast Asia? LocalRealities and Regional Ambiguities. Geografisk Tidsskrift-Danish Journal of Geography 107(1): 29–41.

Roder, W. (2001). Slash-and-Bum Rice Systems in the Hills of North-ern Lao PDR: Description, Challenges, and Opportunities. Inter-national Rice Research Institute (IRRI), Los Baños, Philippines.

Roder, W., Phengchanh, S., and Maniphone, S. (1997). Dynamics ofSoil and Vegetation during Crop and Fallow Period in Slash-and-Burn Fields of Northern Laos. Geoderma 76(1-2): 131–144.

Saito, K., Linquist, B., Keobualapha, B., Shiraiwa, T., and Horie, T.(2006). Farmers’ Knowledge of Soils in Relation to CroppingPractices: A Case Study of Farmers in Upland Rice BasedSlash-and-Burn Systems of Northern Laos. Geoderma 136(1-2):64–74.

Schmidt-Vogt, D., Leisz, S. J., Mertz, O., Heinimann, A., Thiha, T.,Messerli, P., Epprecht, M., Pham, V. C., Vu, K. C., Chi, K.,Hardiono, M., and Dao, T. M. (2009). An Assessment of Trendsin the Extent of Swidden in Southeast Asia. Human Ecology 37:269–280 doi:10.1007/s10745-009-9239-0.

Seidenberg, C., Mertz, O., and Kias, M. B. (2003). Fallow, Labour andLivelihood in Shifting Cultivation: Implications for Deforestationin Northern Lao PDR. Geografisk Tidsskrift-Danish Journal ofGeography 103(2): 71–80.

Thongmanivong, S., and Fujita, Y. (2006). Recent Land Use andLivelihood Transitions in Northern Laos. Mountain Researchand Development 26(3): 237–244.

Thongmanivong, S., Fujita, Y., Phanvilay, K., and Vongvisouk, T.(2009). Agrarian Land Use Transformation in Northern Laos:From Swidden to Rubber. Southeast Asian Studies 47(3): 330–347.

Verburg, P. H., Overmars, K. P., and Witte, N. (2004). Accessibilityand Land-Use Patterns at the Forest Fringe in the North-EasternPart of the Philippines. The Geographical Journal 170: 238–255.

World Bank (2008). Policy, Market, and Agriculture Transition in theNorthern Uplands. World Bank, Washington.

Yamamoto, Y., Oberthür, T., and Lefroy, R. (2009). Spatial Identifica-tion by Satellite Imagery of the Crop–Fallow Rotation Cycle inNorthern Laos. Environment, Development and Sustainability 11(3): 639–654.

Yokoyama, S. (2004). Forest, Ethnicity and Settlement in the Moun-tainous Area of Northern Laos. Southeast Asian Studies 42(2):132–156.

62 Hum Ecol (2013) 41:51–62